0
votes

Using Azure Search REST API v2016-09-01, the following query find the expected document:

?queryType=full&search=id:3119443 AND name:du*
{
  "value": [
    {
      "@search.score": 4.425995,
      "id": "3119443",
      "name": "dupond"
    }
  ]
}

Whereas the following broader query (searching d* instead of du*) does not find the same document:

?queryType=full&search=id:3119443 AND name:d*
{
  "value": []
}

The name field uses a custom analyzer with the Whitespace tokenizer and the WordDelimiterTokenFilter, AsciiFoldingTokenFilter and Lowercase token filters.


  • Most of the indexed documents are correctly found when searching their first name letter.
  • The issue is 100% reproducible on those specific documents, for which I don't find anything special.
  • The Search Service is a "Standard" tier (1 replica, 1 partition, 1 search unit), with the index containing 3,000,000+ documents.

Thank you.

1
Hi, I am Nate from Azure Search. This appears to be a bug. We are looking into the issue. Feel free to contact me at [email protected] if you have any questions.Nate Ko
This is a regression introduced in a recent change. The change introduce a cap on how many terms a wildcard search query can expand to and affects queries that are widely scoped such a* or /.*/. We are treating this as a bug and are actively working on a fix.Nate Ko

1 Answers

1
votes

Thanks for reporting the issue. As commented, this is a regression introduced in a recent change. The bug has been fixed. Thanks.